A novel wavelength shift keying transmitter using a pair of Mach Zehnder modulators

Transcription

1 Available online at Otics Communications 28 (2008) A novel wavelength shift keying transmitter using a air of Mach Zehnder modulators Hao Chi a, *, Xianmin Zhang a, Jianing Yao b a Deartment of Information and Electronic Engineering, Zhejiang University, Hangzhou 30027, China b Microwave Photonics Research Laboratory, School of Information Technology and Engineering, University of Ottawa, Ottawa, ON, Canada KN 6N5 Received 24 October 2007; received in revised form 6 January 2008; acceted 6 January 2008 Abstract A novel and simle method to achieve wavelength shift keying (WSK) modulation is roosed and exerimentally demonstrated. The roosed WSK transmitter consists of a air of Mach Zehnder modulators (MZMs) that are biased aroriately to achieve comlementary oeration. When a non-return-to-zero data stream is alied to the two MZMs, thanks to the comlementary oeration of the MZMs, the WSK-modulated signal can be obtained by combining the two intensity-modulated otical signals. A tunable otical delay line (ODL) is alied to achieve synchronization between the two otical signals. The ODL can also be used to comensate for the disersion-induced walk-off effect between the two otical carriers. The alication of the roosed WSK transmitter in a assive otical network is demonstrated, in which the downstream WSK signal is reused as otical carrier for ustream intensity-modulated data transmission. The roosed aroach is exerimentally realized; exerimental results verify the feasibility of the aroach. Ó 2008 Elsevier B.V. All rights reserved. Keywords: Otical communications; Modulation formats; Wavelength shift keying; Otical frequency shift keying; Mach Zehnder Modulator; Passive otical network. Introduction Wavelength shift keying (WSK) or otical frequency shift keying (OFSK) has been a toic of research interest recently. Due to the feature of constant light intensity, the WSK modulation can be used to reduce the unwanted cross gain modulation in a semiconductor otical amlifier and the four-wave mixing in a WDM system []. It is demonstrated that the WSK modulation can be alied in a assive otical network (PON), where the ustream data is intensity-modulated uon the downstream WSK signal [2,3]. In addition, we may also use the WSK modulation in an otical acket or burst switching network to imlement otical labeling of the ayload data [4]. Among the various WSK formats, binary WSK is the most widely used * Corresonding author. Tel.: address: chihao@zju.edu.cn (H. Chi). modulation formats, which can be imlemented utilizing two otical wavelengths, with one carrying non-return-tozero (NRZ) data stream and the other synchronously carrying the comlementary data stream. At the receiver end, the binary WSK signal can be easily converted to intensitymodulated data by using an otical bandass filter (OBPF) to remove one of the two wavelengths. Several aroaches to imlementing the WSK modulation have been recently roosed. One aroach is based on the conversion from otical hase modulation to intensity modulation by using a delay line interferometer [2]. This aroach can oerate at a high-seed. However, a differentially re-coded data stream should be alied to the hase modulator in this aroach, which increases the comlexity of the system. The WSK modulation has also been demonstrated based on olarization modulation [5,6], in which a high cost olarization modulator is required to achieve comlementary modulation of the /$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:0.06/j.otcom

2 258 H. Chi et al. / Otics Communications 28 (2008) two wavelength channels. Recently, an integrated LiNbO 3 high-seed OFSK modulator based on a secially designed Mach Zehnder structure was demonstrated [7]. In their aroach, the OFSK carrier sacing relies on the inut RF signal, which is limited to the tens of GHz due to the inherent constraint of LiNbO 3. Note that, due to the narrow wavelength sacing between the two wavelength channels, a stable otical filter with narrow assband has to be used to achieve the conversion from WSK modulation to intensity modulation at the receiver side. The OFSK modulation can also be realized based on direct current modulation in a tunable laser; however, the achievable bit rate is limited to several hundreds Mb/s [8]. In this aer, we resent a novel and simle aroach to imlementing high-seed WSK modulation. The roosed WSK transmitter consists of a air of conventional Mach Zehnder modulators (MZMs), which are biased at different transmission oints to achieve comlementary oeration. Thanks to the comlementary oeration of the MZMs, a air of comlementary intensity-modulated otical signal can be generated. By combing the two comlementary signals, the WSK-modulated signal is achieved. In our system the WSK carrier sacing can be continuously tunable and the data rate can be as high as that can be accommodated by the MZMs. To achieve synchronization between the two otical signals, a tunable otical delay line (ODL) is alied in one arm. The ODL can also be used to comensate for the disersion-induced walk-off effect between the two otical carriers. A roof-of-concet exeriment is erformed to demonstrate the alication of the roosed WSK transmitter in PON systems, in which the downstream WSK signal is reused as the otical carrier for ustream data transmission. The resented exerimental results verify the feasibility and otentials of the roosed aroach. 2. Oeration rincile The schematic diagram of the roosed WSK transmitter is illustrated in Fig.. The light beams from two continuous-wave laser diodes (LDs) with small wavelength Fig.. The configuration of the roosed WSK transmitter (PC: olarization controller; MZM: Mach Zehnder modulator; ODL: otical delay line; OC: otical combiner). sacing are fed into two MZMs (MZM and MZM2). Two olarization controllers (PC and PC2) are emloyed at the inuts of the MZMs to align the olarization states of the inut lightwaves with the olarization axis of the MZMs, to minimize the olarization-deendent loss. At the outut of MZM, the PC3 is used to ensure the olarization state of the outut lightwave from MZM is well aligned with that from MZM2 before combining at the otical combiner. A tunable ODL is incororated in the lower arm to ensure synchronization between the two data streams. The NRZ data steam is fed to the two MZMs via a ower divider. The key to achieve a WSK modulation in the roosed scheme is to bias the two MZMs at different transmission oints, to ensure a comlementary oeration of the two MZMs, as shown in the inset of Fig.. Assume that the signal voltage of the inut data stream at bit is V, the bias voltages alied to the two MZMs should be controlled to be V and V V, where V is the half-wave voltage of the MZM (corresonding to the minimum transmission oint). As we know, the normalized transfer function of a singledrive MZM is given as I out ¼ I in 2 þ 2 cos V s þ V bias ; ðþ V where I in and I out are resectively, the inut and outut otical owers, V bias is the bias voltage alied to the MZM, and V s is the voltage of the data stream alied to the MZM, V s ¼ V ; bit : ð2þ 0; bit 0 Therefore, the outut signal from the MZM (with bias voltage V bias ¼ V )is I out k h ¼ I in k ( ¼ i V cos sþv 2 2 þ h V i 2 k cos V V ; bit ; 0; bit 0 and the outut signal at MZM2 (with bias voltage V bias2 ¼ V V )is h i I out k 2 ¼ I in k 2 þ cos V sþv V 2 2 V ( 0; bit ð4þ ¼ h i V cos : ; bit 0 I in 2 k 2 V From (3) and (4), we can see that if the otical owers from the two LDs are equal, the outut data streams from two the MZMs are comlementary. The combination of the two comlementary data streams at the outut of the otical combiner would generate a WSK-modulated signal, given that the otical ath lengths of the two arms are equal (which is ensured by tuning the otical delay line). It is worth noting that the aroach using a air of MZMs that are resectively biased at ositive and negative sloes of the MZM 0 s modulation curve to achieve comlementary ð3þ

3 H. Chi et al. / Otics Communications 28 (2008) modulation has found alications in microwave hotonic filters with negative coefficient(s) [9,0]. 3. Exeriment results and discussion The roosed WSK transmitter as shown in Fig. is exerimentally evaluated. In the exeriment, the wavelengths k and k 2 are set at nm and nm, resectively. The olarization states of the inut lightwaves from the two LDs are adjusted by the two PCs to maximize the outut owers. A 2.5 Gb/s NRZ data stream from a bit error rate tester (BERT) is fed to the two MZMs via a ower divider. The two MZMs are biased at V V and V, where the half-wave voltage V is around 7.9 V and the inut data stream voltage V is around 2.3 V. The ODL in the lower arm is carefully tuned to comensate for the otical ath length difference of the two arms in order to achieve synchronization between the two data streams. PC3 at the outut of MZM is used to ensure the outut lightwave from MZM is well aligned with that from MZM2 before combining at the OC. In addition, the owers of the two LDs are also adjusted to comensate for the loss difference between the two arms in order to ensure the same otical ower levels at the OC. Fig. 2 shows the measured intensity-modulated waveforms at the oututs of MZM and MZM2. The WSK waveform at the outut of the OC is also shown. It is clearly seen that the data streams at the oututs of the two MZMs are comlementary, and the intensity of the combined signal is nearly constant. One of the imortant alications of the roosed WSK modulation technique is in a PON system where the downstream WSK signal can be reused as the otical carrier for ustream data transmission using intensity modulation thanks to the constant envelo of the WSK-modulated signals [2,3]. To demonstrate this alication, a roof-of-concet exeriment with the setu as shown in Fig. 3 is λ λ 2 λ + λ ns/div Fig. 2. The measured intensity-modulated waveforms at the oututs of the two MZMs, and the waveform after combination at the outut of the OC. OLT Data Data ONU Fig. 3. Exerimental setu for demonstrating the reuse of WSK signal as otical carrier for data re-modulation (circ: otical circulator; SSMF: standard single mode fiber; DCF: disersion comensation fiber; OS: otical slitter). imlemented. A air of otical circulators is used to multilex the ustream and downstream signals onto a single otical fiber link. The WSK transmitter is laced at the otical line terminal (OLT) side. The downstream signal from the WSK transmitter, after roagation through 0 km standard single mode fiber (SSMF) and.6 km disersion comensation fiber (DCF), is slit into two beams via a 50:50 otical couler at the otical network unit (ONU) side. One is fed to an otical bandass filter (OBPF) with 0.2 nm bandwidth to select one of the two wavelengths to erform WSK to IM conversion. The converted IM signal is detected by a PIN hoto-receiver. The other beam is reused as the otical carrier which is fed to an MZM to intensity-modulate the ustream data. The ustream data is then transmitted to the OLT through a 0 km SSMF and.6 km DCF. We first investigate the disersion effect on the erformance of the system. To do so, we measure the WSK enveloe after transmission through a length of SSMF. As can be seen in Fig. 4a, the enveloe of the WSK signal is not constant after transmission in a 0 km SSMF due to the disersion-induced walk off between the two wavelength channels. To reduce the walk-off effect, we may choose two wavelengths with smaller wavelength sacing. Fig. 4b and c show the cases where the wavelength sacing are 0. and 0.02 nm, resectively. It is shown that the variations in the enveloe intensity almost disaear in the case when the wavelength sacing is 0.02 nm. However, the use of such small wavelength sacing would necessitate the use of an athermal OBPF with ultra-narrow bandwidth at the receiver end to erform the WSK to IM conversion [3]. An alternative solution is to use disersion comensation fiber (DCF) to eliminate the disersion-induced walk-off while keeing larger wavelength sacing. Fig. 4d shows the WSK enveloe after transmission through 0 km SSMF with disersion comensation using.6 km DCF. A constant intensity enveloe is observed. The bit error rate (BER) erformance of the WSK transmitter is then investigated. In this case, the wavelength sacing is ket as 0.8 nm. We measure the BER vs. the received otical ower for two cases: back-to-back and 0 km SSMF +.6 km DCF transmission. The results are

4 2520 H. Chi et al. / Otics Communications 28 (2008) a b c d 0.5 ns/div Fig. 4. The measured WSK enveloe. (a) Wavelength sacing of 0.8 nm, 0 km SSMF without DCF; (b) wavelength sacing of 0. nm; 0 km SSMF without DCF; (c) wavelength sacing of 0.02 nm, 0 km SSMF without DCF; (d) wavelength sacing of 0.8 nm; 0 km SSMF with.6 km DCF. shown in Fig. 5. A 0.4 db ower enalty is observed at a BER level of 0 9. The sectra of the WSK signal before and after the OBPF are shown in the inset of Fig. 5. Finally, we reuse the received WSK signal to intensitymodulate a 2.5 Gb/s data stream for ustream data transmission. The exerimental setu is also show in Fig. 3. Again two cases are evaluated: back-to-back and 0 km SSMF +.6 km DCF transmission. The BER test results are shown in Fig. 6. A 0.7 db ower enalty is observed at a BER level of 0 9. Although some residual intensity modulation from the downstream WSK channel still remains, considerable erformance of the ustream intensity modulation signal is observed. An eye diagram showing the re-modulated signal for back-to-back transmission when the received ower is 9 dbm, which is given in the inset of Fig. 6. The eye diagram is widely oen which shows a good erformance when reusing the WSK signal for ustream transmission. As mentioned above, the smaller wavelength sacing between the two wavelength channels requires an otical filter with narrower assband; on the other hand, larger wavelength sacing would alleviate this requirement while magnifying the disersion-induced walk-off effect. A disersion comensation scheme is emloyed to suress the disersion-induced walk-off effect in the above exerimental demonstration. Note that a tunable ODL is alied in the roosed WSK transmitter to ensure synchronization of the two comlementary data streams at the outut of the otical combiner. In fact, this ODL can be further emloyed to comensate for the inter-band disersion in the otical link. Therefore, the DCF module is unnecessary if the ODL in the WSK transmitter is finely re-tuned for a given otical link. The BER erformance of the system without DCF module is re-measured in the case the ODL is re-tuned to cancel the inter-band disersion; the results are shown as the triangles in Figs. 5 and 6. Around 0. db ower enalty of the downstream WSK-modulated signal is observed at a BER level of 0 9 comared to the WSK B-B WSK 0 km SSMF+DCF WSK 0 km SSMF+ODL -0 Remodulated IM B-B Remodulated IM 0 km SSMF+DCF Remodulated IM 0 km SSMF+ODL ns/div Fig. 5. Performance of WSK signal transmission. Square: back-to-back; circle: 0 km SSMF with DCF module; triangle: 0 km SSMF with the ODL-based walk-off cancellation (inset: the sectra before and after filtering). Fig. 6. Performance of re-modulated IM signal transmission: Square: back-to-back; circle: 0 km SSMF with DCF module; triangle: 0 km SSMF with the ODL-based walk-off cancellation (inset: the eye diagram of the back-to-back transmission at a received otical ower of 9 dbm).

5 H. Chi et al. / Otics Communications 28 (2008) case with DCF module; and around 0.3 db ower enalty of the ustream intensity-modulated signal is observed. These small ower enalties are mainly attributed to the intra-band disersion. Note that, in a system with higher bit rate or longer otical fiber link, the disersion comensation scheme has to be alied to comensate for the intra-band disersion, since the ODL is only effective for the inter-band disersion. It is worth noting that the basic idea of the disersion canceling based on ODL in our aroach is similar to the aroach roosed in [], in which the ODL is laced at the receiver side to comensate for the grou delay mismatch between the two wavelength channels. The major advantage of the roosed scheme is that a air of MZMs is emloyed to realize WSK modulation without the requirement of secial modulator or data re-coding as comared with the aroaches in [2,5,6]. As comared with the aroach based on direct current modulation in a tunable laser [8], our aroach can suort bit rate as high as tens of Gb/s due to the high bandwidth caacity of MZMs. The LiNbO 3 OFSK modulator roosed in [7] rovided a feasible way to realize highseed OFSK modulation in site of the limitation on the channel sacing. In our exeriment, we have demonstrated that the wavelength sacing in our aroach can be arbitrarily set with the hel of a disersion canceling scheme based on ODL. Larger wavelength sacing would reduce the requirement on the filter in the WSK receiver. Note that in the alication of metroolitan network or access network (which is one of the major alications of the WSK technique), the bandwidth resource in otical fibers is usually not a major limitation. 4. Conclusions We have demonstrated a novel and simle method to achieve WSK modulation. The key in the roosed aroach is the use of two MZMs that are biased at different transmission oints to oerate in a comlementary manner. The WSK signal is obtained by combining the two comlementary data streams. The wavelength sacing between the two carriers of the WSK signal is tunable. In addition, the tunable ODL within the WSK transmitter can be used to comensate for the disersion-induced walk-off effect between the two otical carriers. The WSK transmitter and the reuse of the WSK signal for data remodulation have been exerimentally demonstrated, which verifies the feasibility of the roosed scheme. Acknowledgements This work was suorted artially suorted by the National Natural Science Foundation of China (No ), the Zhejiang Provincial Natural Science Foundation of China (No. Y04073), the Program for New Century Excellent Talents in University (No. NCET-05-58), and the Key Project of Science and Technology Plan of Zhejiang Province of China (No. 2004C2G200007). References [] H.K. Kim, S. Chandrasekhar, IEEE Photon. Technol. Lett. 2 (2000) 42. [2] W. Hung, N. Deng, C.-K. Chan, L.-K. Chen, IEEE Photon. Technol. Lett. 6 (2004) 739. [3] I. Garcés, A. Lóez, J.J. Martínez, A. Villafranca, M.A. Losada, Proc. ECOC 2006, Th [4] L. Xu, T. Wang, O. Matsuda, M. Cvijetic, I. Glesk, P.R. Prucnal, IEEE Photon. Technol. Lett. 7 (2005) [5] S.-S. Pun, C.-K. Chan, L. -K. Chen, IEEE Photon. Technol. Lett. 7 (2005) 528. [6] I.T. Monroy, S.-J. Kim, C. Peucheret, P. Jeesen, Ot. Fiber Technol. 3 (2007) 3. [7] T. Kawanishi, K. Higuma, T. Fujita, J. Ichikawa, T. Sakamoto, S. Shinada, M. Izutsu, Electron. Lett. 40 (2004) 69. [8] V. Polo, J. Prat, J.J.V. Olmos, I.T. Monroy, A.M.J. Koonen, J. Ot. Network. 5 (2006) 739. [9] J. Camany, D. Pastor, A. Martinez, B. Ortega, S. Sales, Ot. Lett. 28 (2003) 45. [0] B. Vidal, J.L. Corral, J. Marti, IEEE Photon. Technol. Lett. 7 (2005) 666. [] T. Kawanishi, T. Fujita, K. Higuma, J. Ichikawa, T. Sakamoto, M. Izutsu, IEICE Electron. Exr. 2 (2005) 333.